compress_scanned_pdf.sh

#!/bin/bash

shopt -s dotglob

# use bash strict mode
set -euo pipefail
IFS=$'\n\t'

if [[ "$#" != "2" ]]; then
    echo "usage: $0 input.pdf output.pdf"
    exit 2
fi

input="$(readlink -f "$1")"
output="$(readlink -f "$2")"

workdir=$(mktemp -d --tmpdir='' compress-pdf-XXXXXXXXX)

cd "${workdir}"

"gs" -o text.pdf -sDEVICE=pdfwrite -dNOPAUSE -dQUIET -dBATCH -dFILTERIMAGE "$input"

pdfimages -j "$input" .

for i in *.jpg; do
    convert -alpha off -auto-threshold otsu -monochrome -quality 100 "$i" "$i.png"
done

img2pdf --output images.pdf *.png 

pdftk text.pdf multistamp images.pdf output "$output"

echo "output written to $output"

rm -rf "$workdir"